set training to validation ratio

8 次查看（过去 30 天）

Ahmed Abdulla 2021-8-24

0
链接

此问题的直接链接

https://ww2.mathworks.cn/matlabcentral/answers/1439899-set-training-to-validation-ratio

回答： Sahil Jain 2021-8-30

I have been using fitrensemble (gradient boosting - LSBoost) to find a model that fits my dataset, however I was wondering if anyone knew how to specify what fraction of the data is for training the machine and what fraction is to validate the model.

2 个评论
显示无隐藏无

KSSV 2021-8-24

General procedure is 70% training, 15% validation and 15% testing.

Ahmed Abdulla 2021-8-24

Thank you for your response, are you aware if this setting can be changed?

请先登录，再进行评论。

请先登录，再回答此问题。

回答（1 个）

Sahil Jain 2021-8-30

0
链接

此回答的直接链接

https://ww2.mathworks.cn/matlabcentral/answers/1439899-set-training-to-validation-ratio#answer_776924

By default, ‘fitrensemble’ does not perform a training/validation split and uses all the provided data as training data. However, multiple cross-validation options are available as name-value pair arguments. These can be found under the Cross-Validation Options on the documentation page of ‘fitrensemble’.